decision

Records evidence-backed go, stop, branch, and finalize decisions for research quests.

Updated Apr 1, 2026
One-click install
npx skills add https://github.com/zniihgnexy/vid_tokenizer --skill decision-zniihgnexy
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: decision
Source: https://github.com/zniihgnexy/vid_tokenizer/tree/main/.ds/codex-home/skills/deepscientist-decision
Command: npx skills add https://github.com/zniihgnexy/vid_tokenizer --skill decision-zniihgnexy

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Long-running research quests often stall or drift when the next step is ambiguous. This Skill forces every consequential transition—continue, branch, reuse a baseline, launch an experiment, write, finalize, reset, or stop—to be made from durable evidence and recorded with a verdict, action, reason, and next direction. ## Core Features & Use Cases - Structured decision records: Captures verdict, canonical action, reason, evidence paths, and next stage for every consequential choice. - Candidate selection discipline: Requires explicit criteria, a named winner, and justification for rejected alternatives when choosing among idea branches, experiment packages, or paper outlines. - User-gated and autonomous modes: Supports blocking decision requests when user preference matters, and self-resolving autonomous routing when the quest contract allows it. - Use Case: After an experiment run finishes with mixed results, use this Skill to weigh the evidence, decide whether to iterate, branch, or move to writing, and durably record the decision with its rationale. ## Quick Start Ask the AI to evaluate the current quest evidence and record a decision on whether to continue, branch, or stop with reasons and next steps.

Frequently Asked Questions about decision

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I decide whether to continue or stop a research quest?▼

State the real question, collect decision-relevant evidence from run artifacts and reports, then choose a verdict (good, bad, neutral, blocked) mapped to an action like continue, reset, or stop. Record the decision with reasons and evidence paths.

How to choose between multiple experiment or idea candidates?▼

List candidate ids, define explicit selection criteria such as evidence quality and expected information gain, name the winner, and explain why alternatives were rejected. Record implementation priority and expected learnings for the chosen package.

When should a decision request block on user input?▼

Use a blocking decision request only when the user must choose before safe continuation, such as preference-sensitive tradeoffs or quest completion approval. In autonomous mode, ordinary route ambiguity is resolved without asking the user.

What evidence sources should research decisions rely on?▼

Decisions should rely on durable evidence: recent run artifacts, report artifacts, baseline state, and quest documents. Memory is only supporting context, and major decisions should never be made from momentum or intuition alone.

When is it wrong to launch an analysis campaign?▼

Analysis campaigns carry extra resource cost and should not be approved casually. Launch one only when the expected information gain clearly justifies the budget, with explicit academic or claim-level value stated in the decision record.